A meta-analysis of the predictive validity of pretrial risk assessment instruments.

OBJECTIVES: Pretrial risk assessments have been widely implemented, yet persistent concerns about their ability to predict violent crime and their accuracy across racial, ethnic, and gender subgroups continue to fuel opposition and, in some jurisdictions, rollbacks of their use. A 2021 systematic review concluded that pretrial risk assessments demonstrated acceptable accuracy in predicting failure to appear and any new arrests overall, but the evidence base at that time was too limited to examine violence prediction specifically or accuracy across subgroups. HYPOTHESES: Our aims were to (a) summarize the current body of evidence examining the predictive validity of pretrial risk assessments and (b) meta-analyze findings regarding predictive validity for failure to appear in court, new criminal arrest, and new violent criminal arrest during the pretrial period, both overall and across subgroups defined by race, ethnicity, and gender. METHOD: We conducted an updated systematic review and meta-analysis of the predictive validity of 15 instruments across 45 samples (43 publications), assessing over 1.6 million pretrial cases. RESULTS: The meta-analyzed predictive validity for total and subscale scores was good for new violent criminal arrest (area under the receiver operating characteristic curve = .665, 95% confidence interval = [.644, .686]), good for any new criminal arrest (area under the receiver operating characteristic curve = .653, 95% confidence interval = [.636, .669]), and fair for failure to appear (area under the receiver operating characteristic curve = .630, 95% confidence interval = [.611, .648]) overall. Differences across gender were small and directionally inconsistent. Differences across race/ethnicity were small, with slightly higher accuracy for White people, though still fair to good accuracy for people of color. CONCLUSIONS: These findings provide the most comprehensive empirical basis for evaluating the opposition to pretrial risk assessment and support the continued, carefully implemented use of instruments. Local validation, limits on judicial overrides, and embedding instruments within broader reform efforts remain essential to realizing their full potential. (PsycInfo Database Record (c) 2026 APA, all rights reserved).